A, Percentages of hypermethylation of CpG island A of the RASSF1A promoter in 3 types of thyroid lesions (follicular thyroid hyperplasia, follicular thyroid adenoma, and follicular thyroid carcinoma) are compared with the percentage of methylation of normal thyroid follicles. Horizontal line in the middle of each box indicates the median; top and bottom borders of the box, 75th and 25th percentiles, respectively; and whiskers, range of data. B, Average of total genome CpG methylation in follicular thyroid hyperplasia samples (n = 24) compared with the average total genome CpG methylation in normal thyroid follicles (n = 10) using enzyme-linked immunosorbent assay–based global methylation analysis (P = .96). Error bars indicate standard deviation.
aP < .001.
Relative messenger RNA expression in hypermethylated samples of follicular thyroid hyperplasia were measured by real-time quantitative polymerase chain reaction and compared with the average expression in normal thyroid tissue (P = .18). Error bar indicates standard deviation.
Hematoxylin-eosin staining (A-D) and immunohistochemical staining for RASSF1A (E-H) and NF-κB (I-L) expression in normal (A, E, and I), follicular thyroid hyperplasia (B, F, and J), follicular thyroid adenoma (C, G, and K), and follicular thyroid carcinoma (D, H, and L) tissue sections (original magnification ×400). Arrowheads indicate representative expression of RASSF1A (E) and NF-κB (J-L).
Brown TC, Juhlin CC, Healy JM, Prasad ML, Korah R, Carling T. Frequent Silencing of RASSF1A via Promoter Methylation in Follicular Thyroid HyperplasiaA Potential Early Epigenetic Susceptibility Event in Thyroid Carcinogenesis. JAMA Surg. 2014;149(11):1146-1152. doi:10.1001/jamasurg.2014.1694
Follicular thyroid hyperplasia (FTH) refers to enlargement of the thyroid gland due to cellular hyperplasia. It is frequently encountered in clinical practice in nontoxic uninodular or multinodular goiter. The genetic and epigenetic events associated with the origin and malignant potential of FTH are poorly understood.
To analyze FTH samples for known recurrent genetic and epigenetic driver events in thyroid neoplasms such as activating mutations in proto-oncogenes BRAF and NRAS and promoter hypermethylation of tumor suppressor genes CDKN2A, PTEN, and RASSF1A.
Design, Setting, and Participants
Clinical characteristics and thyroid specimens were prospectively obtained from 43 patients who underwent thyroid surgery at Yale–New Haven Hospital.
Main Outcomes and Measures
Presence of BRAFV600E and NRAS codon 61 mutations were assessed in FTH. Methylation status of CDKN2A, PTEN, and RASSF1A gene promoters in FTH, follicular thyroid adenoma, and follicular thyroid carcinoma was quantified. Regulation of RASSF1A messenger RNA (mRNA) and protein expression and its potential neoplastic role in FTH were examined.
An exploratory cohort of FTH (n = 10) was negative for BRAFV600E and NRAS codon 61 mutations. In contrast, epigenetic analysis displayed significant promoter hypermethylation of the tumor-suppressor gene RASSF1A in 6 FTH samples (60%) compared with their adjacent normal tissue (P = .01). The overall genome CpG methylation and promoter methylation of PTEN and CDKN2A were unaffected in the lesions. Further analysis of an expanded cohort of patients with FTH (n = 23), follicular thyroid adenoma (n = 10), and follicular thyroid carcinoma (n = 10) showed RASSF1A promoter hypermethylation in 14 (61%), 9 (90%), and 7 (70%), respectively (P < .001). The overall hypermethylation level in FTH showed a statistically significant inverse correlation with RASSF1A mRNA expression (P = .005). Immunohistochemistry demonstrated minimal or no protein expression in most FTH samples studied. To explore the potential neoplastic contribution of RASSF1A downregulation, we analyzed the expression pattern of thyroid proliferation markers Ki-67 and NF-κB in representative samples. Although Ki-67 expression was undetectable, similar to normal tissue, FTH samples expressed high levels of NF-κB, similar to the expression levels in thyroid tumors.
Conclusions and Relevance
We demonstrate silencing of tumor suppressor RASSF1A in a subset of FTH in the absence of other known thyroid cancer–associated genetic and epigenetic changes. Silencing of RASSF1A and concurrent NF-κB activation demonstrate that a subset of FTH shares epigenetic changes and downstream signaling events associated with malignant lesions, suggesting that FTH may have the potential to be a premalignant lesion.
Thyroid cancer is the most common endocrine malignant neoplasm worldwide, accounting for approximately 1% to 5% of cancers in females and 2% of cancers in males.1 Furthermore, multiple studies have demonstrated that the incidence of thyroid cancer is increasing for reasons only partly known.1- 3 Thyroid cancer most commonly originates from thyroid hormone–producing follicular cells. Papillary thyroid cancer is the most common histological type of thyroid cancer, representing 80% to 85% of cases, while follicular thyroid carcinoma (FTC) occurs in 10% to 15% of cases.
Multiple somatic genetic mutations have been demonstrated to play a significant role in thyroid tumorigenesis. Activating mutations of the v-Raf murine sarcoma viral oncogene homolog B1 proto-oncogene (BRAF) and RET-PTC gene rearrangements have been shown to dysregulate mitogen-activated protein kinase pathway signaling and lead to tumorigenesis in papillary thyroid cancer. In contrast, activating mutations of the RAS proto-oncogene family and the chromosomal rearrangement PAX8–peroxisome proliferator-activated receptor γ function to upregulate the phosphatidylinositol 3-kinase–AKT pathway are the primary genetic events associated with FTC.4 Establishing the role of these mutations in thyroid cancer has aided clinicians in the diagnosis, prognosis, and treatment of thyroid cancer.5,6 Despite the prominent role these genetic aberrations play in thyroid tumorigenesis, approximately 35% of differentiated thyroid cancers do not harbor any known genetic alterations associated with tumorigenesis. Furthermore, it remains uncertain whether these genetic mutations are the initiating events leading to thyroid cancer development.4
Multiple epigenetic events, including aberrant gene promoter methylation, have recently been demonstrated to play a significant role in the origin and progression of thyroid cancer. In particular, aberrant methylation of cytosine residues in CpG islands located near or in gene promoters has been shown to silence tumor suppressor gene expression or increase proto-oncogene expression in thyroid cancer.7- 10 Ras-associated domain family 1A (RASSF1A), the most commonly silenced tumor suppressor via promoter hypermethylation in more than 30 cancer types, also has been shown to be epigenetically silenced in thyroid cancer.11 Other frequently silenced tumor suppressor genes hypermethylated in thyroid cancer include cyclin-dependent kinase inhibitor 2A (CDKN2A), a cell cycle regulator, and phosphatase and tensin homolog (PTEN), which antagonizes the phosphatidylinositol 3-kinase–AKT pathway.12- 14
The RASSF1A protein is a ubiquitously expressed 39-kDa tumor suppressor protein of the alternatively spliced RASSF1 gene. The most prominent structural motif of RASSF1A is the Ras association domain, whereby it binds RAS proteins and may alter their function. It also serves as a scaffold for signaling complexes and has been shown in many different cancers to function as a bona fide tumor suppressor. In doing so, RASSF1A affects multiple cellular processes, including microtubule dynamics, maintenance of genomic stability, cell cycle regulation, apoptosis, and cellular motility.11,15 Expression of RASSF1A is regulated, in part, by methylation of RASSF1 CpG island A, a 737–base pair region containing 85 CpG dinucleotides that overlaps the RASSF1A promoter. The RASSF1A promoter has been shown to be methylated in approximately 30% to 70% of thyroid cancer cases, and 2 separate studies have associated gene methylation with decreased gene expression in thyroid cancer or thyroid cancer cell lines.16- 19 Additionally, benign follicular thyroid adenomas (FTAs) have also demonstrated various frequencies (38%-48%) of RASSF1A promoter methylation.19,20 Indeed, RAS proto-oncogene family mutations have been observed in 20% to 25% of FTAs, indicating that these lesions may be a premalignant precursor to FTC.4 Whether RASSF1A silencing is a necessary event in supporting aberrant function of RAS or is an early event in thyroid tumorigenesis is not clear.
The malignant potential and genetic and epigenetic events associated with follicular thyroid hyperplasia (FTH) are poorly understood. Follicular thyroid hyperplasia refers to benign enlargement of the gland due to cellular hyperplasia. It is frequently encountered in clinical practice in cases of nontoxic multinodular goiter and was present in 40% to 50% of thyroid glands in an autopsy study.21 However, FTH may have malignant potential. Hyperplastic thyroid nodules are considered to be of polyclonal origin, typical of most hyperplastic processes, but it has been previously shown that some nodules are monoclonal in origin and growth.22 Furthermore, in a retrospective review of 300 pathological specimens of thyroid hyperplasia, hyperplastic lesions were frequently shown to contain many histological features associated with thyroid cancer, including nuclear and cytological atypia and increased rates of mitosis.21 Moreover, microsatellite instability, a hallmark of genomic instability associated with tumorigenesis, was shown to be present in 35% of cases of multinodular hyperplasia.23 Similarly, several studies have suggested a higher incidence of thyroid cancer in patients with multinodular goiter, although definitive data are lacking.24 These findings suggest a potential premalignant status, at least in a subset of nodules of FTH. Herein, we systematically analyzed FTH to test whether major thyroid cancer–associated genetic and/or epigenetic alterations occur in hyperplastic thyroid nodules.
Patient characteristics including age, sex, lesion size, lesion type, stage, presence or absence of thyroiditis, and thyroid specimens were documented following thyroidectomy. All tissue samples were reviewed by an experienced endocrine pathologist (M.L.P.) before and after sample isolation. Written informed consent was obtained from patients before surgical resection according to a protocol approved by the Yale University Human Investigation Committee. We isolated DNA from fresh frozen samples using the AllPrep DNA/RNA/Protein Kit (Qiagen). Quantity and quality of isolated DNA were assessed by spectrophotometry (NanoDrop Technologies, Inc) and agarose gel electrophoresis, respectively. The BRAFV600E and NRAS codon 61 gene sequences were amplified using BRAF forward primer 5′-TGCTTGCTCTGATAGGAAAATG-3′, BRAF reverse primer 5′-GGCCAAAAATTTAATCAGTGGA-3′, NRAS forward primer 5′-TGAACTTCCCTCCCTCCCTG-3′, and NRAS reverse primer 5′-CGCCTGTCCTCATGTATTGGT-3′, respectively, as previously described.25 Amplicons were visualized and size verified using 1% agarose gel electrophoresis and Sanger sequenced at the Keck DNA Sequencing Facility, Yale University using an Applied Biosystems 3730 capillary instrument and fluorescently labeled dideoxynucleotides (Big Dye Terminator; Life Technologies). All sequences were confirmed with forward and/or reverse primers using Codon Code Aligner software (CodonCode Corp).
The methylation status of CpG islands 114 256, 101 755, and 110 289 of CDKN2A, PTEN, and RASSF1A gene promoters, respectively, was assessed by MethylScreen technology using the Epitect methyl II polymerase chain reaction (PCR) assay (Qiagen) as previously described.26 Briefly, 300 ng of genomic DNA was mock digested or digested with methylation-sensitive or methylation-dependent restriction enzymes individually or together, and the methylation status of the target sequence was measured using real-time quantitative PCR (qPCR) with probes specific to the target promoter sequences (Qiagen). The cycle threshold values were converted into percentages of unmethylated, intermediate-methylated, and hypermethylated CpG values using an algorithm provided by the manufacturer. The percentage of promoter hypermethylation of targeted genes in each lesion was compared with corresponding normal tissue to determine whether promoters of interest were hypermethylated.
Global methylation levels were assessed with enzyme-linked immunosorbent assay–based technology using a MethylFlash Methylated DNA Quantification Kit (Epigentek) as per the manufacturer’s instructions. Briefly, sample DNA and standard DNA were bonded to assay wells for 1.5 hours. Methylated CpG islands were then probed with capture antibody for 30 minutes followed by incubation with detection antibody for 30 minutes. Sample and standard signals were developed for 7 minutes and absorbance was measured at 450 nm using the Promega Glomax Multidetection System. Relative methylation levels were calculated using an algorithm provided by the manufacturer. All assays were performed in duplicate.
We isolated RNA from fresh frozen samples using the AllPrep DNA/RNA/Protein Kit (Qiagen). We used 100 ng of RNA for complementary DNA synthesis using the iScript complementary DNA synthesis kit (Bio-Rad Laboratories). Real-time qPCR was performed on a CFX96 Real-Time System qPCR machine (Bio-Rad Laboratories) in duplicate using TaqMan PCR master mix with primer and probe pairs specific to RASSF1A and housekeeping gene large ribosomal protein (RPLPO). Relative expression levels were calculated using the Livak method (Bio-Rad Laboratories).
Representative 5-µm-thick sections of FTH, FTA, and FTC from formalin-fixed paraffin-embedded tissue samples were selected by an experienced endocrine pathologist (M.L.P.) for study. In total, 8 samples of FTH and FTC and 7 samples of FTA were analyzed. Using standard immunohistochemistry protocols,26 target epitopes were detected using anti-RASSF1A (mouse monoclonal; Abcam), anti–NF-κB (mouse monoclonal; Santa Cruz Biotechnology), or anti–Ki-67 (MIB-1; Dako) primary antibodies followed by corresponding horseradish peroxidase–linked secondary antibodies (all from Santa Cruz Biotechnology) and 3,3′-diaminobenzidine tetrahydrochloride staining (Life Technologies). Sections were counterstained with hematoxylin and mounted using immunohistomount (Santa Cruz Biotechnology).
Continuous variables were assessed for normality of distribution using the Shapiro-Wilk test, then analyzed using a 2-tailed t test for normally distributed variables or the Mann-Whitney U test for variables with a skewed distribution. For categorical variables, Pearson χ2 test or Fisher exact test was used as appropriate. Correlation between hypermethylation of RASSF1A and fold expression changes by real-time qPCR for RASSF1A in hypermethylated FTH samples was analyzed by bivariate correlation using Pearson test and linear regression. All data analyses were performed using SPSS version 19.0 statistical software (IBM).
We initially evaluated 10 FTH samples for thyroid cancer–associated genetic and epigenetic markers. All 10 samples of FTH were assessed for BRAFV600E and NRAS codon 61 gene mutations and both mutations were absent (Table 1). We then quantified the methylation levels of the promoters of RASSF1A, PTEN, and CDKN2A in the same 10 samples as well as in corresponding normal tissue. Because the highest level of methylation noted in any tested tumor suppressor gene promoter of normal thyroid tissue was 5.2%, with an average methylation level of 1.6%, lesions with more than 10% of their CpG islands methylated were designated as hypermethylated. Using this stringent criteria, 6 samples of FTH were hypermethylated at the RASSF1A promoter with an average methylation level of 24.1% observed. In contrast, hypermethylation was not observed in the promoters of PTEN or CDKN2A in any FTH sample (P = .01) (Table 1). We then analyzed an extended cohort to assess RASSF1A promoter hypermethylation in additional samples. In total, 23 samples of FTH, 10 samples of FTA, and 10 samples of FTC were analyzed for RASSF1A promoter hypermethylation. The demographic characteristics, lesion characteristics, and clinical data of the cohort are shown in Table 2. To assess the significance of the methylation results in this expanded cohort, we compared them with an aggregate cohort of normal thyroid tissue from 29 samples. The average level of RASSF1A promoter methylation in normal thyroid tissues was 1.7% and no normal sample demonstrated evidence of RASSF1A promoter hypermethylation. In contrast, RASSF1A promoter hypermethylation was observed in 14 FTH samples (61%). Similarly, 9 FTA samples (90%) and 7 FTC samples (70%) were found to be hypermethylated, respectively (P < .001) (Figure 1A). To test whether the observed increase in RASSF1A promoter methylation is part of the global methylation changes in the FTH genome, we used a total genome CpG methylation assay and the total levels of CpG methylation were found to be very similar between normal thyroid tissue and FTH (P = .96) (Figure 1B).
Using real-time qPCR and immunohistochemistry techniques, we tested whether promoter hypermethylation is associated with reduced expression of RASSF1A. Although not statistically significant, the overall expression of RASSF1A in hypermethylated samples of FTH was found to be 18% less than the average expression of RASSF1A in normal thyroid tissue (P = .18) (Figure 2). Furthermore, hypermethylation levels were found to inversely correlate with RASSF1A messenger RNA expression using Pearson test and linear regression (P = .005). Samples of FTH, FTA, FTC, and normal thyroid were then examined by immunohistochemistry to assess for RASSF1A protein expression. Minimal or no RASSF1A protein expression was observed in most FTH, FTA, and FTC samples (Figure 3F-H), while RASSF1A protein expression was found in normal adjacent thyroid tissue (Figure 3E). To test whether promoter hypermethylation and reduced expression of RASSF1A potentiates neoplastic changes in FTH, we analyzed the expression pattern of Ki-67 and NF-κB in FTH by immunohistochemistry, 2 proliferation markers frequently found activated in FTA and FTC. Increased expression of the proliferation marker Ki-67 has been shown to be associated with malignant thyroid cancer.27 No FTHs examined showed any detectable increase from basal levels of Ki-67 expression, concurring with their benign phenotype. On the other hand, NF-κB, a neoplastic marker that has been shown to be active in FTA and FTC,28 was overexpressed in the cytoplasm of FTH but not in adjacent normal thyroid tissue (Figure 3I-L).
It is not clear whether thyroid hyperplasia possesses malignant potential. Subsets of hyperplastic nodules have been shown to contain many of the histological features associated with thyroid cancer, can be monoclonal in origin, and demonstrate microsatellite instability.21- 23 Furthermore, several studies have suggested a higher incidence of thyroid cancer in the setting of hyperplastic nodules, although definitive evidence is lacking.24,29,30 Thus, the role hyperplastic nodules may have in thyroid tumorigenesis and their preneoplastic potential remain uncertain.
Multiple studies have suggested a potential role of aberrant DNA methylation in the origin and/or progression of thyroid cancer.7- 9 In this study, we examined the role of epigenetic regulation of RASSF1A, one the most frequently silenced tumor suppressors via promoter methylation, in thyroid tumorigenesis.18- 20 Hypermethylation of the RASSF1A promoter has been found to frequently occur in FTC as well as FTA, suggesting a critical role in aberrant RASSF1A promoter hypermethylation in thyroid tumorigenesis.
This study demonstrates that epigenetic silencing of RASSF1A could be one of the first events in thyroid tumorigenesis because other known cancer-associated genetic and epigenetic aberrations were not identified in FTH. No BRAFV600E or NRAS codon 61 gene mutations were detected in FTH, demonstrating that these somatic mutations that are frequently found in thyroid cancer may not occur in FTH and possibly represent later events in thyroid tumorigenesis. Similarly, promoter hypermethylation of CDKN2A and PTEN was not observed in FTH.
The absence of global methylation changes also suggests that RASSF1A promoter methylation may occur as a targeted methylation event with a potential driving effect on the early stages of thyroid tumorigenesis. Methylation of tumor suppressor genes, including RASSF1A, in later stages of thyroid neoplasia usually occurs in the setting of genome-wide changes in methylation patterns.31,32 As such, methylation of any single tumor suppressor gene in thyroid cancer may occur as a passenger event due to global epigenetic dysregulation. We also examined the expression profiles of Ki-67 and NF-κB, 2 proliferation markers frequently found activated in FTAs and FTCs, to determine whether RASSF1A silencing is associated with signaling events linked to thyroid tumorigenesis. We detected no Ki-67 protein expression, suggesting the absence of a dysregulated proliferation pattern that is characteristic of malignancy. In contrast, NF-κB, a downstream target of a plethora of transformation-promoting signaling events, was found to be markedly upregulated in FTH compared with normal tissue. A previous study demonstrated that NF-κB is upregulated and has an antiapoptotic effect in thyroid cancer.33 These findings suggest that RASSF1A silencing may promote tumorigenesis, in part through upregulation of NF-κB expression. However, the cytoplasmic localization of NF-κB observed in the FTH samples also suggests the need for additional signaling events to activate its transcription-promoting effect. It is unclear at this point whether NF-κB expression is a cause or effect of RASSF1A downregulation. Alternatively, signaling events that promote the FTH phenotype may be activating these events independent of each other.
We demonstrate that RASSF1A promoter hypermethylation and associated silencing of RASSF1A gene expression occur frequently in FTH. While we did not test for all known thyroid genetic and epigenetic aberrations, the most common and prominent thyroid cancer–associated genetic mutations were not detected in FTH, suggesting that RASSF1A promoter hypermethylation may be an early event in thyroid tumorigenesis that potentially causes predisposition of FTH to thyroid adenoma and carcinoma formation.
Corresponding Author: Tobias Carling, MD, PhD, Yale Endocrine Neoplasia Laboratory, Department of Surgery, Yale University School of Medicine, 333 Cedar St, TMP FMB130A, PO Box 208062, New Haven, CT 06520 (firstname.lastname@example.org).
Accepted for Publication: May 27, 2014.
Published Online: September 17, 2014. doi:10.1001/jamasurg.2014.1694.
Author Contributions: Drs Brown and Korah had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Brown, Juhlin, Healy, Korah, Carling.
Acquisition, analysis, or interpretation of data: Brown, Juhlin, Healy, Prasad.
Drafting of the manuscript: Brown, Carling.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Brown, Healy.
Obtained funding: Carling.
Administrative, technical, or material support: Carling.
Study supervision: Juhlin, Prasad, Korah, Carling.
Conflict of Interest Disclosures: None reported.
Funding/Support: Dr Juhlin was supported by a clinical postdoctoral appointment from the Stockholm County Council. Dr Carling is a Damon Runyon Clinical Investigator supported in part by the Damon Runyon Cancer Research Foundation.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Previous Presentation: This paper was presented at the 2014 Annual Meeting of the Association of VA Surgeons; April 6, 2014; New Haven, Connecticut.